[NLPL Task Force (A)] software environment on Abel

[Stephan Oepen]

colleagues,

i realize there is a bit of an underground tradition of do-it-yourself
singularity containers in our group, but in my view there are also
downsides to this approach to consider: containerization creates
obstacles to interoperability and reuse (with others, as well as
across different HPC environments).  also, the singularity layer adds
additional complexity, often unnecessarily so.

the NLPL project is working to create a collection of ready-to-use,
up-to-date, interoperable modules on Abel (and the finnish Taito
supercluster, where there is additional cpu and gpu capacity).  the
goal is to provide everything that would be required for common MSc or
doctoral projects in NLP.  for some background on what is available to
date, please see:

http://wiki.nlpl.eu/index.php/Infrastructure/software/catalogue

if you find something missing or not working according to your
expectations, we will be happy to hear from you.  the technical point
of contact for NLPL is its infrastructure task force (of which i am a
member): ‘infrastructure at nlpl.eu’.

with many new group members, i have been thinking it might make sense
to hold a technical orientation on computing infrastructure at some
point, maybe early january?  for example, a walk-through of sofware
and data under the NLPL umbrella and collective exchange of best
practises and recipes for running effectively on Abel?

best, oe